This study presents a modeling framework for quantifying human impacts and for partitioning the sources of contamination related to water quality in the mixed-use landscape of a small tropical volcanic island. On Tutuila, the main island of American Samoa, production wells in the most populated region (the Tafuna-Leone Plain) produce most of the island’s drinking water. However, much of this water has been deemed unsafe to drink since 2009. Tutuila has three predominant anthropogenic non-point-groundwater-pollution sources of concern: on-site disposal systems (OSDS), agricultural chemicals, and pig manure. These sources are broadly distributed throughout the landscape and are located near many drinking-water wells. Water quality analyses show a link between elevated levels of total dissolved groundwater nitrogen (TN) and areas with high non-point-source pollution density, suggesting that TN can be used as a tracer of groundwater contamination from these sources. The modeling framework used in this study integrates land-use information, hydrological data, and water quality analyses with nitrogen loading and transport models. The approach utilizes a numerical groundwater flow model, a nitrogen-loading model, and a multi-species contaminant transport model. Nitrogen from each source is modeled as an independent component in order to trace the impact from individual land-use activities. Model results are calibrated and validated with dissolved groundwater TN concentrations and inorganic δ¹⁵N values, respectively. Results indicate that OSDS contribute significantly more TN to Tutuila’s aquifers than other sources, and thus should be prioritized in future water-quality management efforts.